- Email: [email protected]
Biomechanical evaluation of pedicle screws versus pedicle and laminar hooks in the thoracic spine
The Spine Journal 6 (2006) 444–449
Case Report
Biomechanical evaluation of pedicle screws versus pedicle and laminar hooks in the thoracic spine Andrew Cordista, MDa, Bryan Conrad, MEnga,*, MaryBeth Horodyski, EdDa, Sheri Walters, MSb, Glenn Rechtine, MDc a
Department of Orthopaedics and Rehabilitation, University of Florida, 1600 SW Archer Road, Gainesville, FL 32610, USA b Department of Physical Therapy, University of Florida, 1600 SW Archer Road, Gainesville, FL 32610, USA c Department of Orthopaedics, University of Rochester, 601 Elmwood Avenue, Rochester, NY 14642, USA Received 21 April 2005; accepted 22 August 2005
Abstract
BACKGROUND CONTEXT: Pedicle screws have been shown to be superior to hooks in the lumbar spine, but few studies have addressed their use in the thoracic spine. PURPOSE: The objective of this study was to biomechanically evaluate the pullout strength of pedicle screws in the thoracic spine and compare them to laminar hooks. STUDY DESING/SETTING: Twelve vertebrae (T1–T12) were harvested from each of five embalmed human cadavers (n560). The age of the donors averaged 83þ8.5 years. After bone mineral density had been measured in the vertebrae (mean50.47 g/cm3), spines were disarticulated. Some pedicles were damaged during disarticulation or preparation for testing, so that 100 out of a possible 120 pullout tests were performed. METHODS: Each vertebra was secured using a custom-made jig, and a posteriorly directed force was applied to either the screw or the claw. Constructs were ramped to failure at 3 mm/min using a Mini Bionix II materials testing machine (MTS, Eden Prairie, MN). RESULTS: Pedicle claws had an average pullout strength of 577 N, whereas the pullout strength of pedicle screws averaged 309 N. Hooks installed using the claw method in the thoracic spine had an overwhelming advantage in pullout strength versus pedicle screws. Even in extremely osteoporotic bone, the claw withstood 88% greater pullout load. CONCLUSION: The results of this study indicate that hooks should be considered when supplemental instrumentation is required in thoracic vertebrae, especially in osteoporotic bone. Ó 2006 Elsevier Inc. All rights reserved.
Keywords:
Thoracic spine; Fusion; Internal fixation; Pedicle screws
Introduction Pedicle screws have been studied widely in the thoracolumbar and lumbar spine and have been shown to be clinically superior to hooks [1–3]. However, few studies have addressed their use in the thoracic spine. Heller et al. [4]
FDA device/drug status: approved for this indication (Synthes pedicle screw system; Moss Miami pedicle and laminar hook system). This work was supported by the Department of Orthopaedics and Rehabilitation, University of Florida. Nothing of value received from a commercial entity related to this manuscript. * Corresponding author. Department of Orthopaedics and Rehabilitation, PO Box 100246, Gainesville, FL 32610. Tel.: (352) 392-7570; fax: (352) 392-8637. E-mail address: [email protected] (B. Conrad) 1529-9430/06/$ – see front matter Ó 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.spinee.2005.08.015
compared pedicle screws with transverse process screws in the upper thoracic spine. Also in the thoracic spine, Dvorak et al. [5] compared conventional pedicle screw fixation with extrapedicular screw fixation. Limited use of thoracic pedicle screws may be due to the increased difficulty in placement, particularly in the midthoracic vertebrae (secondary to small concave pedicles) and the increased risk of spinal cord injury [6]. While evaluating the internal and external morphology of vertebral pedicles, Moran et al. [7] demonstrated that adequate bone stock was available at T2, T7, T12, and L1–L5 spinal levels to accept screws in the 4- to 7-mm diameter range. This conclusion was based on the direct dimensional measurements particularly focusing on the angle the pedicle makes with the vertebral body: the minimum cross-sectional
A. Cordista et al. / The Spine Journal 6 (2006) 444–449
445
dimensions of the pedicle and the distance through the pedicle into the body. The authors reported that it is possible to fit rather large diameter (7 mm) screws into the pedicles of the lumbar vertebrae; however, difficulties arise if these are used in the thoracic vertebrae, especially in the midthoracic region. When investigating the morphometric parameters, Heller et al. and Cinotti et al. [4,8] reported a general trend in the pedicles from T1 to T4 of decreased width and increased height and length. It was also reported that the medial-lateral dimension shrank until pedicles become more elliptical in cross section, furthering the challenge for screw insertion in the upper thoracic pedicles. To date few studies have directly compared fixation with pedicle screws versus laminar and pedicle hooks in the thoracic spine, and there is no clear consensus on which fixation is most appropriate biomechanical standpoint. The primary purpose of this investigation was to determine whether screws or hooks provided the greater resistance to pullout in the thoracic spine. Our secondary aim was to evaluate the effect of bone mineral density (BMD) on the pullout force of thoracic pedicle screws and hooks. Methods Because of the lack of availability of fresh frozen specimens at our institution, we chose to use formalin-fixed cadaver tissue for this investigation. With the approval of the State Anatomical Board of Florida, five formalin-fixed human cadaveric thoracic spines (T1–T12) were harvested and stored at –20 C. The average age of the donors was 838.5 years. To account for variations in bone density and to rule out spines with existing neoplasm or fractures, lateral spine dual-energy X-ray absorptiometry (DEXA) (GE Lunar DPX-L, Madison, WI) scans were taken for four of five specimens before testing. DEXA scans were performed by placing the intact thoracic spines in a bed of rice to simulate the surrounding soft tissues of the trunk. After measuring BMD, all soft tissue was carefully removed and each vertebra was disarticulated. Standard 6.5-mm titanium pedicle screws (Synthes USS, Paoli, PA) and stainless steel pedicle and laminar hooks (Moss Miami, DePuy Spine Inc., Raynham, MA) were used for testing. Screws were inserted under direct visual control after first entering the pedicle with a rongeur. Screw angulation at each level was adjusted to accommodate for anatomical variation in pedicle geometry to avoid cortical penetration. At each level, the pedicle screw was inserted to one-half of the total vertebral anterior-posterior width. Pedicle and laminar hooks were applied in the claw construct, with a pedicle hook inferior and a laminar hook superior (Fig. 1). Screws were tested against claw constructs on the contralateral side of the vertebra in alternating (right, left) fashion. Each vertebra was secured using a custommade jig which allowed the bone to self-align to the direction in which force was applied (Fig. 2). A 12.5-mm aluminum rod was passed through the spinal canal of each
Fig. 1. Synthes USS screw and DePuy Moss Miami claw constructs used for testing.
vertebrae and through two oversized brackets secured to the base of the testing machine. This fixture allowed the vertebral body to rotate in the transverse plane about the rod. Care was taken to minimize the bending moment that was introduced when the rod was not directly in line with the axis of the pedicle screw. Using an MTS Mini Bionix 858 material testing machine (MTS Systems, Eden Prairie, MN), a posterior directed force in the frontal plane was applied to either the screw or the claw. An appropriate diameter rod was attached to both the screw and claw constructs to facilitate attachment to the MTS. Constructs were ramped to failure in displacement control at 3 mm/min while load, displacement, and time data were collected at a sampling rate of 10 Hz. Pullout force was defined as the maximum resistance of the implant to the applied load. From the mechanical testing data, load-displacement curves were generated and peak load values were calculated. A univariate analysis of variance (SPSS, Inc., Chicago, IL) with level and treatment as fixed factors was used to test for significant differences in peak load. To validate the use of embalmed spines, a separate sample of three fresh frozen human vertebrae (T10–T12) was collected and tested for comparison. Screws and claws were applied to the unembalmed spines and tested using the same protocol as described above for the embalmed samples.
Results Eight of 60 vertebrae were excluded because of damage during disarticulation or preparation for testing. Six of
446
A. Cordista et al. / The Spine Journal 6 (2006) 444–449
Fig. 2. Testing setup for axial loading. The vertebra is secured in the testing fixtures by a 10-mm rod through the spinal canal, and tensile force was applied through a universal joint, allowing the vertebra to self-align in the direction loading.
these vertebra were in the T4–T8 region, where the pedicles are very small. A single pedicle was damaged during testing in four specimens (3 screw sides, 1 claw side), making direct comparison impossible for those vertebrae. In all, pullout tests were completed for 83% of the pedicles (100 of 120). Test data was available for both treatments in 48 of 60 vertebrae (80%). The average age of the cadavers was 838.5 years, and the vertebrae were osteoporotic. BMD was assessed for 4 of
the 5 cadavers (48 of 60 specimens). BMD was not measured on vertebrae from the first cadaver because the DEXA machine had not been purchased at the time the study began. The T-score for these vertebrae averaged 3.23, which indicates that BMD was more than 3 standard deviations below the normal value for a healthy 30-yearold. BMD was averaged for each vertebral level, and results were analyzed using a single-factor analysis of variance. Significant differences in BMD by vertebral level were not found (p5.21, Fig. 3). Overall, BMD of the retrieved spinal specimens averaged 0.450.09 g/cm3. When all levels of the thoracic spine are considered, the average pullout strength of the claw construct was significantly higher than the screws, 577248 N and 309189 N respectively (p5.009, Fig. 4). There were no significant differences in pullout force at the different vertebral levels of the spine (p5.679), neither was there an interaction between vertebral level and treatment with pullout force (p5.709). Typical load-displacement curves for a screw construct and a claw construct are presented in Figures 5 and 6, respectively. To determine if BMD was a factor in construct strength, pullout force was analyzed with univariate analysis of variance by defining high BMD (0.45 g/cm3 and higher) and low BMD (below 0.45 g/cm3). No trends were observed between BMD and pullout strength (p5.771), nor were there any interactions between treatment and BMD (p5.207). Mode of failure differed between groups. Screws tend to fail by stripping out the bone contained in the threads of the screw. However, claws typically failed by fracture of the contralateral pedicle or lamina. Although the pullout strength of both implants was greater in the unembalmed bone, the ratio of claw/screw pullout forces was similar in fresh frozen and embalmed pedicles. The average pullout strength for the 49 screws successfully tested in embalmed vertebrae was 313.33 N. The average pullout force required was 592 N for the 51 claws tested in pedicles from embalmed vertebrae. In the fresh-frozen vertebrae, the pullout strength of the three claws and three screws averaged 1073 N and 512 N, respectively. This yielded a ratio of claw/screw of 1.9 for embalmed vertebrae and 2.1 for fresh-frozen. Discussion In the osteoporotic bone we tested, we found the claw had a much higher pullout force compared with the screw. This appears logical because screw fixation is probably more affected by decreased density in the cancellous bone than the hooks, which are attached to cortical bone. Cortical bone is affected by osteoporosis later in life and therefore would potentially have a smaller negative impact on hook fixation. Somewhat surprising was the fact that the fixation with the pedicle screw was not related to the size of the pedicle. We expected fixation at the lower thoracic
A. Cordista et al. / The Spine Journal 6 (2006) 444–449
447
0.8 0.7 0.6
BMD
0.5 0.4 0.3 0.2 0.1 0 1
2
3
4
5
6
7
8
9
10
11
12
Vertebral Level Fig. 3. Mean bone mineral density (BMD) by vertebral level. There are no significant differences between levels.
spine to be better because of the larger pedicles. In order to see such a difference, it might have been necessary to use a larger screw in the lower vertebrae. Cinotti et al. [8] and others have cautioned that pedicles between T4 and T8 might not be wide enough to accommodate screw fixation. Our findings were in agreement with this. BMD was lowest in vertebrae T4–T8, and this was also the region where vertebrae were damaged most often in disarticulation or testing. Use of thoracic pedicle screws is controversial [7, 9–11], in part because of concerns about the close proximity of the screws to neural structures. There are also differences of opinion on the best method of insertion. The two most popular techniques are sagittal trajectory of the screw so that it parallels the superior end plate of the vertebral body and a trajectory in which the screw parallels the anatomic axis of the pedicle. The issue of the proper placement of a thoracic pedicle screw, whether it be transpedicular or extrapedicular or into the transverse processes is currently undecided [4,12,13]. Because of the difficulties involved in inserting thoracic pedicle screws, alternative solutions have been sought,
including the use of hooks in the thoracic spine [1,2,9,10]. To date few studies have directly compared fixation with pedicle screws versus laminar and pedicle hooks in the thoracic spine, and no clear consensus exists on which fixation is superior. In a recent study, Hackenberg et al. [14] found that pullout strength of pedicle screws was superior to hooks in healthy spines; however, in osteoporotic spines (!100 mg/mL) they found no difference between screws and hooks. In a similar study by the same group, Liljenqvist et al. [15] reported that the average pullout strength was significantly greater for the pedicle screws both in the upper and lower thoracic spine in comparison to the pedicle and laminar hooks. Both Hackenberg [14] and Liljenqvist [15] used only a single hook in their experiments. Clinically, a single hook would not be used; therefore, we feel that to make a true comparison between screws and hooks a claw design must be employed [16,17]. Another recent study by Gayet et al. [18] examined the forces required to extract pedicle screws and claw constructs from cadaver thoracic vertebrae. Gayet et al., who used a two-level claw technique, concluded that 4-mm screws are 23% weaker than hooks, while 5-mm screws
900
300 250
700 600
*
Force (N)
Pullout Force (N)
800
500 400 300 200
200 Failure Load
150 100 50
100 0
CLAW
SCREW
Fig. 4. Mean pullout force for all levels for claw versus screw constructs. *p5.008
0
0
0.5
1
1.5
2
2.5
3
Displacement (mm) Fig. 5. Typical load-displacement curve for screw construct.
3.5
448
A. Cordista et al. / The Spine Journal 6 (2006) 444–449 700 600
Force (N)
500 Failure Load
400 300 200 100 0
0
1
2
3
4
5
6
7
8
Displacement (mm) Fig. 6. Typical load-displacement curve for claw construct.
are only 12% stronger. However, they found that screws were most likely to pull out because the bony anchoring fails. This led them to the conclusion that screws are less effective if anchoring cannot be guaranteed, which is a common problem in osteoporotic bone. In a study of four types of posterior fixation in the lumbar spine, Coe et al. [2] reported that laminar hooks were clearly superior in strength to single screw fixation in osteoporotic spine. Using an artificial model, Margulies et al. [19] concluded that claw hooks are suitable alternatives to pedicle screws. Our results indicating that hooks are superior to screws in resisting pullout compare well with those of Gayet et al. [18], Coe et al. [2], and Margulies et al. [19]. In addition to in vitro biomechanical studies, several clinical studies [20,21] have reported clinical outcomes in patients receiving pedicle screws. In a comparative study by Sasso and Cotler [20], 70 patients were treated for unstable fractures and fracture-dislocations of the thoracic and lumbar spinal regions with three fixation devices, including Harrington rods and hooks. The authors contend that not only have the hooks demonstrated similar sagittal plane corrections and return of deformity in comparison to the screws in the thoracic spine and thoracolumbar junction, but that the hooks also subsequently required only one-half the operative time and blood loss. This also supports the application of hooks for injuries in which the posterior column is intact. In addition, the Harrington hooks were shown to be superior in preventing collapse of the vertebral body at the thoracolumbar junction during 12 months follow-up. In a study evaluating instrumentation in the treatment of idiopathic scoliosis, Wimmer et al. [21] found no difference between correction of thoracic and lumbar curves when using constructs consisting of pedicle screws versus laminar hooks. There are several limitations to our study. Because fresh cadaver tissue was unavailable, we chose to make use of embalmed cadaver tissue. Although there is speculation that the embalming process has some effect on the mechanical properties of bone [20,22–28], previous research on the subject is equivocal. In addition, both treatments would be equally affected and therefore a valid comparison between the relative values of each implant could be made.
Researchers have presented conflicting reports on the effects of formalin on the mechanical properties of skeletal tissue. As early as 1964, McElhaney et al. [23] reported only a small decrease in compressive strength (12%), and no change in tensile strength of cortical bone due to formaldehyde fixation. Likewise, Sedlin and Hirsch [24] found no change in bending modulus in specimens that had been fixed in formalin for 3 weeks. In a study using sheep vertebrae, Edmondston et al. [25] found a slight increase in compressive strength as a result of formalin fixation. However, they found no differences in either BMD or BMC. This result is supported by previous research performed by Blanton and Biggs [26] who found that embalming did not affect the density of bone. Currey et al. [27] evaluated bovine cortical bone in both bending, tensile, and impact tests. They concluded that formalin fixation had no effect on bending or tensile strength or Young’s modulus but did result in reduced energy absorption in impact. From mechanical tests performed on the humeri and femora of cats, Goh et al. [28] determined that strength and stiffness of the bones were unaffected by formalin fixation; however, preserved bones did demonstrate a marked decreased in energy absorbed during failure. In summary, it seems that the strength and stiffness of bone is not altered in a predictable manner. Any changes in mechanical properties are likely to be small. It is possible that if cancellous bone is more adversely affected by embalming than cortical bone, claw strength would be preferentially biased because screws rely on cancellous bone for purchase. However, we also confirmed with a series of fresh specimens that at least the same ratio of pullout strength between screw and claw was maintained in the embalmed specimens as in the fresh specimens. Eight vertebrae were destroyed in the process of either disarticulation or testing. In some cases, the bone was of very poor quality and damage occurred during the effort to insert a knife at the facets and separate the vertebral bodies. Others were lost when the claw side was tested first and failed, which tended to cause a catastrophic fracture of the contralateral pedicle. Thus, in later testing, the screw side was tested first. Although it is possible that testing of the pedicle screw first could affect the performance of the hooks adversely, screw pullout did not produce the contralateral damage seen when the claw failed. Any effect the testing order might have would result in an underestimation of the actual hook strength, which would not have changed the results of the study. Clinically, the damage to the bone associated with screw stripping might be preferable to the contralateral fractures associated with claw failure; however, the much higher failure strength of the claw construct makes failure less likely. Our testing evaluated only one aspect of the loading (pullout) that implants experience in clinical use. We believe that pullout resistance is a critical factor because proximal to the apex of the kyphosis, particularly in a kyphotic
A. Cordista et al. / The Spine Journal 6 (2006) 444–449
deformity, the implant is being loaded in direct pullout [29]. Further research should be directed toward evaluating the response of pedicle screws and hooks to different loading conditions. Our results indicate that hooks have much better pullout performance in osteoporotic bone. Although pedicle screws have been shown to be excellent in the lumbar spine, the technical challenges and risk of neurological complications also make their use less attractive in the thoracic spine. We are not advocating that pedicle screws be abandoned as a means of fixation in the thoracic spine, rather that the surgeon’s choice of instrumentation must be appropriate to satisfy the individual patient’s biomechanical needs. For some applications, including correction of translation and rotational deformity, the pedicle screw may be the best option. This study demonstrates that pedicle hooks definitely should be considered when supplemental instrumentation is required in thoracic vertebrae, especially in bone that is osteoporotic. Acknowledgments The authors thank Joanne Clarke for editorial assistance and Randy Zinnerman, Amber Martin, and Dupree Hatch for technical assistance. References [1] Barr SJ, Schuette AM, Emans JB. Lumbar pedicle screws versus hooks: results in double major curves in adolescent idiopathic scoliosis. Spine 1997;22:1369–79. [2] Coe JD, Warden KE, Herzig MA, et al. Influence of bone mineral density on the fixation of thoracolumbar implants: a comparative study of transpedicular screws, laminar hooks, and spinous process wires. Spine 1990;15:902–7. [3] Steffee AD, Biscup RS, Sitkowski DJ. Segmental spine plates with pedicle screw fixation: a new internal fixation device for disorders of the lumbar and thoracolumbar spine. Clin Orthop 1986;203:45–53. [4] Heller JG, Shuster JK, Hutton WC. Pedicle and transverse process screws of the upper thoracic spine: biomechanical comparison of loads to failure. Spine 1999;24:654–8. [5] Dvorak M, MacDonald S, Gurr KR, et al. An anatomic, radiographic, and biomechanical assessment of extrapedicular screw fixation in the thoracic spine. Spine 1993;18:1689–94. [6] Liljenqvist UR, Link TM, Halm HF. Morphometric analysis of thoracic and lumbar vertebrae in idiopathic scoliosis. Spine 2000;25:1247–53. [7] Moran JM, Berg WS, Berry JL, et al. Transpedicular screw fixation. J Orthop Res 1989;7:107–14. [8] Cinotti G, Gumina S, Ripani M, Postacchini F. Pedicle instrumentation in the thoracic spine: a morphometric and cadaveric study for placement of screws. Spine 1999;24:114–9. [9] Cohen-Gadol AA, Dekutoski MB, Kim CW, Quasi LM, Krauss WE. Safety of supplemental endplate screws in thoracic pedicle hook fixation. J Neurosurg 2003;29(1 Suppl.):31–5.
449
[10] Lehman RA Jr, Polly DW Jr, Kuklo TR, Cunningham B, Kirk KL, Belmont PJ Jr. Straight-forward versus anatomic trajectory technique of thoracic pedicle screw fixation: a biomechanical analysis. Spine 2003;28:2058–65. [11] Lehman RA Jr, Kuklo TR. Use of anatomic trajectory for thoracic pedicle screw salvage after failure/violation using the straightforward technique: a biomechanical analysis. Spine 2003;28:2072–7. [12] Halvorson TL, Kelley LA, Thomas KA, Whitecloud TS, Cook SD. Effects of bone mineral density on pedicle screw fixation. Spine 1994;19:2415–20. [13] Mac-Thiong JM, Labelle H, Aubin CE. Thoracic pedicle screw insertion using a transpedicular drill guide: a preliminary study. J Spinal Disord Tech 2004;17:29–32. [14] Hackenberg L, Link T, Liljenqvist U. Axial and tangential fixation strength of pedicle screws versus hooks in the thoracic spine in relation to bone mineral density. Spine 2002;27:937–42. [15] Liljenqvist U, Hackenberg L, Link T, et al. Pullout strength of pedicle screws versus pedicle and laminar hooks in the thoracic spine. Acta Orthop Belg 2001;67:157–63. [16] Benzel EC. Short-segment compression instrumentation for selected thoracic and lumbar spine fractures: the short-rod/two-claw technique. J Neurosurg 1993;79:335–40. [17] Roach JW, Ashman RB, Allard RN. The strength of a posterior element claw at one versus two spinal levels. J Spinal Disord 1990; 3:259–61. [18] Gayet LE, Pries P, Hamcha H, et al. Biomechanical study and digital modeling of traction resistance in posterior thoracic implants. Spine 2002;27:707–14. [19] Margulies JY, Caruso SA, Chattar-Cora D, et al. Substitution of transpedicular screws by hook claws in a vertebrectomy model. J Spinal Disord 1998;11:36–40. [20] Sasso RC, Cotler HB. Posterior instrumentation and fusion for unstable fractures and fracture-dislocations of the thoracic and lumbar spine: a comparative study of three fixation devices in 70 patients. Spine 1993;18:450–60. [21] Wimmer C, Gluch H, Nogler M, et al. Treatment of idiopathic scoliosis with CD-instrumentation: lumbar pedicle screws versus laminar hooks in 66 patients. Acta Orthop Scand 2001;72:615–20. [22] Wilke HJ, Krischak S, Claes LE. Formalin fixation strongly influences biomechanical properties of the spine. J Biomechanics 1996;29: 1629–31. [23] McElhaney J, Fogle J, Byars E, et al. Effect of embalming on the mechanical properties of beef bone. J Appl Physiol 1964;19: 1234–6. [24] Sedlin ED, Hirsch C. Factors affecting the determination of the physical properties of femoral cortical bone. Acta Orthop Scand 1966;37: 29–48. [25] Edmondston SJ, Singer KP, Day RE, et al. Formalin fixation effects on vertebral bone density and failure mechanics: an in-vitro study of human and sheep vertebrae. Clin Biomech 1994;9:175–9. [26] Blanton PL, Biggs NL. Density of fresh and embalmed human compact and cancellous bone. Am J Phys Anthropol 1968;29:39–44. [27] Currey JD, Brear K, Zioupos P, et al. Effect of formaldehyde fixation on some mechanical properties of bovine bone. Biomaterials 1995; 16:1267–71. [28] Goh JCH, Ang EJ, Bose K. Effect of preservation medium on the mechanical properties of cat bones. Acta Orthop Scand 1989;60:465–7. [29] Ashman RB, Galpin RD, Corin JD, Johnston CE 2nd. Biomechanical analysis of pedicle screw instrumentation systems in a corpectomy model. Spine 1989;14:1398–405.
Case Report
Biomechanical evaluation of pedicle screws versus pedicle and laminar hooks in the thoracic spine Andrew Cordista, MDa, Bryan Conrad, MEnga,*, MaryBeth Horodyski, EdDa, Sheri Walters, MSb, Glenn Rechtine, MDc a
Department of Orthopaedics and Rehabilitation, University of Florida, 1600 SW Archer Road, Gainesville, FL 32610, USA b Department of Physical Therapy, University of Florida, 1600 SW Archer Road, Gainesville, FL 32610, USA c Department of Orthopaedics, University of Rochester, 601 Elmwood Avenue, Rochester, NY 14642, USA Received 21 April 2005; accepted 22 August 2005
Abstract
BACKGROUND CONTEXT: Pedicle screws have been shown to be superior to hooks in the lumbar spine, but few studies have addressed their use in the thoracic spine. PURPOSE: The objective of this study was to biomechanically evaluate the pullout strength of pedicle screws in the thoracic spine and compare them to laminar hooks. STUDY DESING/SETTING: Twelve vertebrae (T1–T12) were harvested from each of five embalmed human cadavers (n560). The age of the donors averaged 83þ8.5 years. After bone mineral density had been measured in the vertebrae (mean50.47 g/cm3), spines were disarticulated. Some pedicles were damaged during disarticulation or preparation for testing, so that 100 out of a possible 120 pullout tests were performed. METHODS: Each vertebra was secured using a custom-made jig, and a posteriorly directed force was applied to either the screw or the claw. Constructs were ramped to failure at 3 mm/min using a Mini Bionix II materials testing machine (MTS, Eden Prairie, MN). RESULTS: Pedicle claws had an average pullout strength of 577 N, whereas the pullout strength of pedicle screws averaged 309 N. Hooks installed using the claw method in the thoracic spine had an overwhelming advantage in pullout strength versus pedicle screws. Even in extremely osteoporotic bone, the claw withstood 88% greater pullout load. CONCLUSION: The results of this study indicate that hooks should be considered when supplemental instrumentation is required in thoracic vertebrae, especially in osteoporotic bone. Ó 2006 Elsevier Inc. All rights reserved.
Keywords:
Thoracic spine; Fusion; Internal fixation; Pedicle screws
Introduction Pedicle screws have been studied widely in the thoracolumbar and lumbar spine and have been shown to be clinically superior to hooks [1–3]. However, few studies have addressed their use in the thoracic spine. Heller et al. [4]
FDA device/drug status: approved for this indication (Synthes pedicle screw system; Moss Miami pedicle and laminar hook system). This work was supported by the Department of Orthopaedics and Rehabilitation, University of Florida. Nothing of value received from a commercial entity related to this manuscript. * Corresponding author. Department of Orthopaedics and Rehabilitation, PO Box 100246, Gainesville, FL 32610. Tel.: (352) 392-7570; fax: (352) 392-8637. E-mail address: [email protected] (B. Conrad) 1529-9430/06/$ – see front matter Ó 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.spinee.2005.08.015
compared pedicle screws with transverse process screws in the upper thoracic spine. Also in the thoracic spine, Dvorak et al. [5] compared conventional pedicle screw fixation with extrapedicular screw fixation. Limited use of thoracic pedicle screws may be due to the increased difficulty in placement, particularly in the midthoracic vertebrae (secondary to small concave pedicles) and the increased risk of spinal cord injury [6]. While evaluating the internal and external morphology of vertebral pedicles, Moran et al. [7] demonstrated that adequate bone stock was available at T2, T7, T12, and L1–L5 spinal levels to accept screws in the 4- to 7-mm diameter range. This conclusion was based on the direct dimensional measurements particularly focusing on the angle the pedicle makes with the vertebral body: the minimum cross-sectional
A. Cordista et al. / The Spine Journal 6 (2006) 444–449
445
dimensions of the pedicle and the distance through the pedicle into the body. The authors reported that it is possible to fit rather large diameter (7 mm) screws into the pedicles of the lumbar vertebrae; however, difficulties arise if these are used in the thoracic vertebrae, especially in the midthoracic region. When investigating the morphometric parameters, Heller et al. and Cinotti et al. [4,8] reported a general trend in the pedicles from T1 to T4 of decreased width and increased height and length. It was also reported that the medial-lateral dimension shrank until pedicles become more elliptical in cross section, furthering the challenge for screw insertion in the upper thoracic pedicles. To date few studies have directly compared fixation with pedicle screws versus laminar and pedicle hooks in the thoracic spine, and there is no clear consensus on which fixation is most appropriate biomechanical standpoint. The primary purpose of this investigation was to determine whether screws or hooks provided the greater resistance to pullout in the thoracic spine. Our secondary aim was to evaluate the effect of bone mineral density (BMD) on the pullout force of thoracic pedicle screws and hooks. Methods Because of the lack of availability of fresh frozen specimens at our institution, we chose to use formalin-fixed cadaver tissue for this investigation. With the approval of the State Anatomical Board of Florida, five formalin-fixed human cadaveric thoracic spines (T1–T12) were harvested and stored at –20 C. The average age of the donors was 838.5 years. To account for variations in bone density and to rule out spines with existing neoplasm or fractures, lateral spine dual-energy X-ray absorptiometry (DEXA) (GE Lunar DPX-L, Madison, WI) scans were taken for four of five specimens before testing. DEXA scans were performed by placing the intact thoracic spines in a bed of rice to simulate the surrounding soft tissues of the trunk. After measuring BMD, all soft tissue was carefully removed and each vertebra was disarticulated. Standard 6.5-mm titanium pedicle screws (Synthes USS, Paoli, PA) and stainless steel pedicle and laminar hooks (Moss Miami, DePuy Spine Inc., Raynham, MA) were used for testing. Screws were inserted under direct visual control after first entering the pedicle with a rongeur. Screw angulation at each level was adjusted to accommodate for anatomical variation in pedicle geometry to avoid cortical penetration. At each level, the pedicle screw was inserted to one-half of the total vertebral anterior-posterior width. Pedicle and laminar hooks were applied in the claw construct, with a pedicle hook inferior and a laminar hook superior (Fig. 1). Screws were tested against claw constructs on the contralateral side of the vertebra in alternating (right, left) fashion. Each vertebra was secured using a custommade jig which allowed the bone to self-align to the direction in which force was applied (Fig. 2). A 12.5-mm aluminum rod was passed through the spinal canal of each
Fig. 1. Synthes USS screw and DePuy Moss Miami claw constructs used for testing.
vertebrae and through two oversized brackets secured to the base of the testing machine. This fixture allowed the vertebral body to rotate in the transverse plane about the rod. Care was taken to minimize the bending moment that was introduced when the rod was not directly in line with the axis of the pedicle screw. Using an MTS Mini Bionix 858 material testing machine (MTS Systems, Eden Prairie, MN), a posterior directed force in the frontal plane was applied to either the screw or the claw. An appropriate diameter rod was attached to both the screw and claw constructs to facilitate attachment to the MTS. Constructs were ramped to failure in displacement control at 3 mm/min while load, displacement, and time data were collected at a sampling rate of 10 Hz. Pullout force was defined as the maximum resistance of the implant to the applied load. From the mechanical testing data, load-displacement curves were generated and peak load values were calculated. A univariate analysis of variance (SPSS, Inc., Chicago, IL) with level and treatment as fixed factors was used to test for significant differences in peak load. To validate the use of embalmed spines, a separate sample of three fresh frozen human vertebrae (T10–T12) was collected and tested for comparison. Screws and claws were applied to the unembalmed spines and tested using the same protocol as described above for the embalmed samples.
Results Eight of 60 vertebrae were excluded because of damage during disarticulation or preparation for testing. Six of
446
A. Cordista et al. / The Spine Journal 6 (2006) 444–449
Fig. 2. Testing setup for axial loading. The vertebra is secured in the testing fixtures by a 10-mm rod through the spinal canal, and tensile force was applied through a universal joint, allowing the vertebra to self-align in the direction loading.
these vertebra were in the T4–T8 region, where the pedicles are very small. A single pedicle was damaged during testing in four specimens (3 screw sides, 1 claw side), making direct comparison impossible for those vertebrae. In all, pullout tests were completed for 83% of the pedicles (100 of 120). Test data was available for both treatments in 48 of 60 vertebrae (80%). The average age of the cadavers was 838.5 years, and the vertebrae were osteoporotic. BMD was assessed for 4 of
the 5 cadavers (48 of 60 specimens). BMD was not measured on vertebrae from the first cadaver because the DEXA machine had not been purchased at the time the study began. The T-score for these vertebrae averaged 3.23, which indicates that BMD was more than 3 standard deviations below the normal value for a healthy 30-yearold. BMD was averaged for each vertebral level, and results were analyzed using a single-factor analysis of variance. Significant differences in BMD by vertebral level were not found (p5.21, Fig. 3). Overall, BMD of the retrieved spinal specimens averaged 0.450.09 g/cm3. When all levels of the thoracic spine are considered, the average pullout strength of the claw construct was significantly higher than the screws, 577248 N and 309189 N respectively (p5.009, Fig. 4). There were no significant differences in pullout force at the different vertebral levels of the spine (p5.679), neither was there an interaction between vertebral level and treatment with pullout force (p5.709). Typical load-displacement curves for a screw construct and a claw construct are presented in Figures 5 and 6, respectively. To determine if BMD was a factor in construct strength, pullout force was analyzed with univariate analysis of variance by defining high BMD (0.45 g/cm3 and higher) and low BMD (below 0.45 g/cm3). No trends were observed between BMD and pullout strength (p5.771), nor were there any interactions between treatment and BMD (p5.207). Mode of failure differed between groups. Screws tend to fail by stripping out the bone contained in the threads of the screw. However, claws typically failed by fracture of the contralateral pedicle or lamina. Although the pullout strength of both implants was greater in the unembalmed bone, the ratio of claw/screw pullout forces was similar in fresh frozen and embalmed pedicles. The average pullout strength for the 49 screws successfully tested in embalmed vertebrae was 313.33 N. The average pullout force required was 592 N for the 51 claws tested in pedicles from embalmed vertebrae. In the fresh-frozen vertebrae, the pullout strength of the three claws and three screws averaged 1073 N and 512 N, respectively. This yielded a ratio of claw/screw of 1.9 for embalmed vertebrae and 2.1 for fresh-frozen. Discussion In the osteoporotic bone we tested, we found the claw had a much higher pullout force compared with the screw. This appears logical because screw fixation is probably more affected by decreased density in the cancellous bone than the hooks, which are attached to cortical bone. Cortical bone is affected by osteoporosis later in life and therefore would potentially have a smaller negative impact on hook fixation. Somewhat surprising was the fact that the fixation with the pedicle screw was not related to the size of the pedicle. We expected fixation at the lower thoracic
A. Cordista et al. / The Spine Journal 6 (2006) 444–449
447
0.8 0.7 0.6
BMD
0.5 0.4 0.3 0.2 0.1 0 1
2
3
4
5
6
7
8
9
10
11
12
Vertebral Level Fig. 3. Mean bone mineral density (BMD) by vertebral level. There are no significant differences between levels.
spine to be better because of the larger pedicles. In order to see such a difference, it might have been necessary to use a larger screw in the lower vertebrae. Cinotti et al. [8] and others have cautioned that pedicles between T4 and T8 might not be wide enough to accommodate screw fixation. Our findings were in agreement with this. BMD was lowest in vertebrae T4–T8, and this was also the region where vertebrae were damaged most often in disarticulation or testing. Use of thoracic pedicle screws is controversial [7, 9–11], in part because of concerns about the close proximity of the screws to neural structures. There are also differences of opinion on the best method of insertion. The two most popular techniques are sagittal trajectory of the screw so that it parallels the superior end plate of the vertebral body and a trajectory in which the screw parallels the anatomic axis of the pedicle. The issue of the proper placement of a thoracic pedicle screw, whether it be transpedicular or extrapedicular or into the transverse processes is currently undecided [4,12,13]. Because of the difficulties involved in inserting thoracic pedicle screws, alternative solutions have been sought,
including the use of hooks in the thoracic spine [1,2,9,10]. To date few studies have directly compared fixation with pedicle screws versus laminar and pedicle hooks in the thoracic spine, and no clear consensus exists on which fixation is superior. In a recent study, Hackenberg et al. [14] found that pullout strength of pedicle screws was superior to hooks in healthy spines; however, in osteoporotic spines (!100 mg/mL) they found no difference between screws and hooks. In a similar study by the same group, Liljenqvist et al. [15] reported that the average pullout strength was significantly greater for the pedicle screws both in the upper and lower thoracic spine in comparison to the pedicle and laminar hooks. Both Hackenberg [14] and Liljenqvist [15] used only a single hook in their experiments. Clinically, a single hook would not be used; therefore, we feel that to make a true comparison between screws and hooks a claw design must be employed [16,17]. Another recent study by Gayet et al. [18] examined the forces required to extract pedicle screws and claw constructs from cadaver thoracic vertebrae. Gayet et al., who used a two-level claw technique, concluded that 4-mm screws are 23% weaker than hooks, while 5-mm screws
900
300 250
700 600
*
Force (N)
Pullout Force (N)
800
500 400 300 200
200 Failure Load
150 100 50
100 0
CLAW
SCREW
Fig. 4. Mean pullout force for all levels for claw versus screw constructs. *p5.008
0
0
0.5
1
1.5
2
2.5
3
Displacement (mm) Fig. 5. Typical load-displacement curve for screw construct.
3.5
448
A. Cordista et al. / The Spine Journal 6 (2006) 444–449 700 600
Force (N)
500 Failure Load
400 300 200 100 0
0
1
2
3
4
5
6
7
8
Displacement (mm) Fig. 6. Typical load-displacement curve for claw construct.
are only 12% stronger. However, they found that screws were most likely to pull out because the bony anchoring fails. This led them to the conclusion that screws are less effective if anchoring cannot be guaranteed, which is a common problem in osteoporotic bone. In a study of four types of posterior fixation in the lumbar spine, Coe et al. [2] reported that laminar hooks were clearly superior in strength to single screw fixation in osteoporotic spine. Using an artificial model, Margulies et al. [19] concluded that claw hooks are suitable alternatives to pedicle screws. Our results indicating that hooks are superior to screws in resisting pullout compare well with those of Gayet et al. [18], Coe et al. [2], and Margulies et al. [19]. In addition to in vitro biomechanical studies, several clinical studies [20,21] have reported clinical outcomes in patients receiving pedicle screws. In a comparative study by Sasso and Cotler [20], 70 patients were treated for unstable fractures and fracture-dislocations of the thoracic and lumbar spinal regions with three fixation devices, including Harrington rods and hooks. The authors contend that not only have the hooks demonstrated similar sagittal plane corrections and return of deformity in comparison to the screws in the thoracic spine and thoracolumbar junction, but that the hooks also subsequently required only one-half the operative time and blood loss. This also supports the application of hooks for injuries in which the posterior column is intact. In addition, the Harrington hooks were shown to be superior in preventing collapse of the vertebral body at the thoracolumbar junction during 12 months follow-up. In a study evaluating instrumentation in the treatment of idiopathic scoliosis, Wimmer et al. [21] found no difference between correction of thoracic and lumbar curves when using constructs consisting of pedicle screws versus laminar hooks. There are several limitations to our study. Because fresh cadaver tissue was unavailable, we chose to make use of embalmed cadaver tissue. Although there is speculation that the embalming process has some effect on the mechanical properties of bone [20,22–28], previous research on the subject is equivocal. In addition, both treatments would be equally affected and therefore a valid comparison between the relative values of each implant could be made.
Researchers have presented conflicting reports on the effects of formalin on the mechanical properties of skeletal tissue. As early as 1964, McElhaney et al. [23] reported only a small decrease in compressive strength (12%), and no change in tensile strength of cortical bone due to formaldehyde fixation. Likewise, Sedlin and Hirsch [24] found no change in bending modulus in specimens that had been fixed in formalin for 3 weeks. In a study using sheep vertebrae, Edmondston et al. [25] found a slight increase in compressive strength as a result of formalin fixation. However, they found no differences in either BMD or BMC. This result is supported by previous research performed by Blanton and Biggs [26] who found that embalming did not affect the density of bone. Currey et al. [27] evaluated bovine cortical bone in both bending, tensile, and impact tests. They concluded that formalin fixation had no effect on bending or tensile strength or Young’s modulus but did result in reduced energy absorption in impact. From mechanical tests performed on the humeri and femora of cats, Goh et al. [28] determined that strength and stiffness of the bones were unaffected by formalin fixation; however, preserved bones did demonstrate a marked decreased in energy absorbed during failure. In summary, it seems that the strength and stiffness of bone is not altered in a predictable manner. Any changes in mechanical properties are likely to be small. It is possible that if cancellous bone is more adversely affected by embalming than cortical bone, claw strength would be preferentially biased because screws rely on cancellous bone for purchase. However, we also confirmed with a series of fresh specimens that at least the same ratio of pullout strength between screw and claw was maintained in the embalmed specimens as in the fresh specimens. Eight vertebrae were destroyed in the process of either disarticulation or testing. In some cases, the bone was of very poor quality and damage occurred during the effort to insert a knife at the facets and separate the vertebral bodies. Others were lost when the claw side was tested first and failed, which tended to cause a catastrophic fracture of the contralateral pedicle. Thus, in later testing, the screw side was tested first. Although it is possible that testing of the pedicle screw first could affect the performance of the hooks adversely, screw pullout did not produce the contralateral damage seen when the claw failed. Any effect the testing order might have would result in an underestimation of the actual hook strength, which would not have changed the results of the study. Clinically, the damage to the bone associated with screw stripping might be preferable to the contralateral fractures associated with claw failure; however, the much higher failure strength of the claw construct makes failure less likely. Our testing evaluated only one aspect of the loading (pullout) that implants experience in clinical use. We believe that pullout resistance is a critical factor because proximal to the apex of the kyphosis, particularly in a kyphotic
A. Cordista et al. / The Spine Journal 6 (2006) 444–449
deformity, the implant is being loaded in direct pullout [29]. Further research should be directed toward evaluating the response of pedicle screws and hooks to different loading conditions. Our results indicate that hooks have much better pullout performance in osteoporotic bone. Although pedicle screws have been shown to be excellent in the lumbar spine, the technical challenges and risk of neurological complications also make their use less attractive in the thoracic spine. We are not advocating that pedicle screws be abandoned as a means of fixation in the thoracic spine, rather that the surgeon’s choice of instrumentation must be appropriate to satisfy the individual patient’s biomechanical needs. For some applications, including correction of translation and rotational deformity, the pedicle screw may be the best option. This study demonstrates that pedicle hooks definitely should be considered when supplemental instrumentation is required in thoracic vertebrae, especially in bone that is osteoporotic. Acknowledgments The authors thank Joanne Clarke for editorial assistance and Randy Zinnerman, Amber Martin, and Dupree Hatch for technical assistance. References [1] Barr SJ, Schuette AM, Emans JB. Lumbar pedicle screws versus hooks: results in double major curves in adolescent idiopathic scoliosis. Spine 1997;22:1369–79. [2] Coe JD, Warden KE, Herzig MA, et al. Influence of bone mineral density on the fixation of thoracolumbar implants: a comparative study of transpedicular screws, laminar hooks, and spinous process wires. Spine 1990;15:902–7. [3] Steffee AD, Biscup RS, Sitkowski DJ. Segmental spine plates with pedicle screw fixation: a new internal fixation device for disorders of the lumbar and thoracolumbar spine. Clin Orthop 1986;203:45–53. [4] Heller JG, Shuster JK, Hutton WC. Pedicle and transverse process screws of the upper thoracic spine: biomechanical comparison of loads to failure. Spine 1999;24:654–8. [5] Dvorak M, MacDonald S, Gurr KR, et al. An anatomic, radiographic, and biomechanical assessment of extrapedicular screw fixation in the thoracic spine. Spine 1993;18:1689–94. [6] Liljenqvist UR, Link TM, Halm HF. Morphometric analysis of thoracic and lumbar vertebrae in idiopathic scoliosis. Spine 2000;25:1247–53. [7] Moran JM, Berg WS, Berry JL, et al. Transpedicular screw fixation. J Orthop Res 1989;7:107–14. [8] Cinotti G, Gumina S, Ripani M, Postacchini F. Pedicle instrumentation in the thoracic spine: a morphometric and cadaveric study for placement of screws. Spine 1999;24:114–9. [9] Cohen-Gadol AA, Dekutoski MB, Kim CW, Quasi LM, Krauss WE. Safety of supplemental endplate screws in thoracic pedicle hook fixation. J Neurosurg 2003;29(1 Suppl.):31–5.
449
[10] Lehman RA Jr, Polly DW Jr, Kuklo TR, Cunningham B, Kirk KL, Belmont PJ Jr. Straight-forward versus anatomic trajectory technique of thoracic pedicle screw fixation: a biomechanical analysis. Spine 2003;28:2058–65. [11] Lehman RA Jr, Kuklo TR. Use of anatomic trajectory for thoracic pedicle screw salvage after failure/violation using the straightforward technique: a biomechanical analysis. Spine 2003;28:2072–7. [12] Halvorson TL, Kelley LA, Thomas KA, Whitecloud TS, Cook SD. Effects of bone mineral density on pedicle screw fixation. Spine 1994;19:2415–20. [13] Mac-Thiong JM, Labelle H, Aubin CE. Thoracic pedicle screw insertion using a transpedicular drill guide: a preliminary study. J Spinal Disord Tech 2004;17:29–32. [14] Hackenberg L, Link T, Liljenqvist U. Axial and tangential fixation strength of pedicle screws versus hooks in the thoracic spine in relation to bone mineral density. Spine 2002;27:937–42. [15] Liljenqvist U, Hackenberg L, Link T, et al. Pullout strength of pedicle screws versus pedicle and laminar hooks in the thoracic spine. Acta Orthop Belg 2001;67:157–63. [16] Benzel EC. Short-segment compression instrumentation for selected thoracic and lumbar spine fractures: the short-rod/two-claw technique. J Neurosurg 1993;79:335–40. [17] Roach JW, Ashman RB, Allard RN. The strength of a posterior element claw at one versus two spinal levels. J Spinal Disord 1990; 3:259–61. [18] Gayet LE, Pries P, Hamcha H, et al. Biomechanical study and digital modeling of traction resistance in posterior thoracic implants. Spine 2002;27:707–14. [19] Margulies JY, Caruso SA, Chattar-Cora D, et al. Substitution of transpedicular screws by hook claws in a vertebrectomy model. J Spinal Disord 1998;11:36–40. [20] Sasso RC, Cotler HB. Posterior instrumentation and fusion for unstable fractures and fracture-dislocations of the thoracic and lumbar spine: a comparative study of three fixation devices in 70 patients. Spine 1993;18:450–60. [21] Wimmer C, Gluch H, Nogler M, et al. Treatment of idiopathic scoliosis with CD-instrumentation: lumbar pedicle screws versus laminar hooks in 66 patients. Acta Orthop Scand 2001;72:615–20. [22] Wilke HJ, Krischak S, Claes LE. Formalin fixation strongly influences biomechanical properties of the spine. J Biomechanics 1996;29: 1629–31. [23] McElhaney J, Fogle J, Byars E, et al. Effect of embalming on the mechanical properties of beef bone. J Appl Physiol 1964;19: 1234–6. [24] Sedlin ED, Hirsch C. Factors affecting the determination of the physical properties of femoral cortical bone. Acta Orthop Scand 1966;37: 29–48. [25] Edmondston SJ, Singer KP, Day RE, et al. Formalin fixation effects on vertebral bone density and failure mechanics: an in-vitro study of human and sheep vertebrae. Clin Biomech 1994;9:175–9. [26] Blanton PL, Biggs NL. Density of fresh and embalmed human compact and cancellous bone. Am J Phys Anthropol 1968;29:39–44. [27] Currey JD, Brear K, Zioupos P, et al. Effect of formaldehyde fixation on some mechanical properties of bovine bone. Biomaterials 1995; 16:1267–71. [28] Goh JCH, Ang EJ, Bose K. Effect of preservation medium on the mechanical properties of cat bones. Acta Orthop Scand 1989;60:465–7. [29] Ashman RB, Galpin RD, Corin JD, Johnston CE 2nd. Biomechanical analysis of pedicle screw instrumentation systems in a corpectomy model. Spine 1989;14:1398–405.